Abstract: During the operation of the pebble bed high-temperature gas-cooled reactors (HTGRs), the wear of fuel elements will lead to the generation of nonspherical graphite dust. Graphite dust migrates and diffuses in the primary loop, which has adverse effects on the performance and maintenance of the device. The drag coefficient is the significant parameter to determine particle movement characteristics. In order to develop the correlation of aerodynamic drag of nonspherical graphite dust particles in HTGRs, computational fluid dynamics was used to calculate the drag coefficient of particles based on the SST kω turbulent model. The simulation method was verified by using the empirical drag coefficient of spherical particles. The geometric structure of nonspherical particles was modeled by the combined sphere method according to graphite dust morphology sampled by the AVR reactor. The components of aerodynamic drag including pressure drag and shear drag for microscale graphite dust and the drag variation due to different types of particles and inflow angles were discussed. The results show that particle Reynolds number and inflow angle are important parameters to determine particle drag coefficient. The drag coefficient decreases with the increase of particle Reynolds number while increases with the increase of inflow angle. The shear drag is larger than the pressure drag in most cases which causes the drag coefficient of long rodlike particles higher. The shear drag is mainly affected by particle Reynolds number while the pressure drag is mainly affected by inflow angle. Based on the drag coefficients of several typical nonspherical particles, a dimensionless nonspherical characteristic parameter was derived by the ratio of particle geometry projection parameters. Then, a drag coefficient correlation for nonspherical graphite dust was developed and the dimensionless characteristic parameter was used to reflect the effects of particle morphology and inflow angle on the drag coefficient. Compared with the existing correlations of nonspherical drag coefficient, the correlation proposed in this study depends on a unique dimensionless characteristic parameter and can predict the effect of the particle Reynolds number and inflow angle on drag coefficient with low deviation. To obtain a more general drag coefficient model of graphite particles, assuming that the morphology and inflow angle of graphite particles obey uniform probability density distribution, then, a statisticalaverage drag coefficient model suitable for graphite particles was established and verified by drag coefficient measurement data from a freefall experiment of graphite dust. The results show that they have high consistency. Therefore, the present numerical prediction method, proposed empirical correlation and statisticalaverage model can provide a basis for the study of the motion characteristics of graphite dust in HTGRs.